You are on page 1of 15

Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

Heat and Mass Transfer in a Clay-Pot Refrigerator: Analysis


Revisited

R.M. Damle and A.W. Date*


Department of Mechanical Engineering
Indian Institute of Technology Bombay
Powai, Mumbai 400 076
India

Abstract

In this paper, steady-state performance of a clay-pot refrigerator which works on the


evaporative cooling principle, is analysed using Reynolds Flow Model of convective
heat/mass transfer including effects of radiation. For the assumed respiratory cooling load
of the preserved foods, the preservation temperature is predicted under a variety of
ambient temperatures and relative humidities. The paper also corrects for the oversight-
error introduced in the previous published paper [1].

Keywords—Reynolds Flow Model, Heat/Mass Transfer, Clay-pot Refrigerator.

INTRODUCTION
In rural areas of India, vegetarian food is often preserved in a clay-pot
refrigerator. The cooling space is a smaller clay pot inserted within a larger clay
pot (see fig 1 - left). The annular space between the two pots is filled with sand
or tiny pebbles. The voids in the annular space are occupied by water. Convective
and radiative heat transfer Qin from the hot and dry surroundings evaporates this
water and brings about cooling of the space in the inner pot where food is
kept. Once the sensible heats are removed, the preserved foods give out
1
respiratory cooling load qload (typically 0.05 ~ 0.2 W/kg) . Thus, the Steady-State
Coefficient of Performance (COP) of such a refrigerator may be defined as

Qload qload  m food


COP   (1)
Qin Qin

This definition is of course different from that used for a conventional


refrigerator in which Qin is replaced by Work input Win. But, whereas one pays
for Win, Q in is free. Also, Win of a conventional refrigerator is under designer’s
control whereas Qin depends on ambient conditions and, therefore, not under
designer’s control.
*
Corresponding author.
Email address: awdate@iitb.ac.in (A.W. Date)
1
In addition, there may be cooling load due to heat transfers from the top and bottom of the refrigerator.

©Regional Centre for Energy, Heat and Mass Transfer for Asia and the Pacific/Printed in India
12 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

NOMENCLATURE

A surface area (m2) λ latent heat (J/kg)


B Spalding number ( - ) Φ relative humidity
Bi Biot number ( - )  viscosity (N-s/m2)
b clay wall thickness variable, (m)
Cp specific heat (J/kg-K) Suffixes
g mass transfer coefficient (kg/m2 -s) a air or atmospheric
H pot height (m) cl clay
h Enthalpy of mixture (J/kg) cold inner pot environment
k Thermal conductivity (W/m-K) dp dew point
kp permeability (m2 ) e energy conservation principle
khy hydraulic conductivity (m/s) eff sand + water
Le Lewis number ( - ) i inner pot
ṁ evaporation rate (kg/s) in radiation + Convection
m mass (kg) load cooling load
p pressure (N/m2 ) m mass conservation principle or mean
Q heat transfer (W) M transferred substance state
T temperature (o C) mean mixed mean
nc natural convection
Greek symbols o outer pot
α heat transfer coefficient (W/m2 - K) rad radiation
β volumetric coefficient K-1 ref reference value
ω mass fraction v vapour
ηth thermodynamic efficiency w water or w-w state
 emissivity ∞ ambient condition

In order to circumvent the above difficulty, here, an efficiency th is newly defined
based on thermodynamic considerations as [1]

 (2)

where the preservation temperature Tcold in the inner pot depends on the ambient
conditions viz. temperature T and relative humidity  as well as irreversible
heat and mass transfer processes. On the other hand, if the moist ambient air was
cooled at constant pressure then the moisture will begin to condense at what is
called the dew point temperature T dp . We may now imagine that our
refrigerator is used to bring about this ideal cooling. Then, th = 1 will represent
maximum realisable cooling performance.
Figure 1 (right) shows the assumed model of the refrigerator. The curved
surfaces are replaced by straight cylindrical surfaces of radii ri and r0 with height
H. Inner and outer clay wall thicknesses are bi(<< ri) and b0(<<r0) respectively.
The thermal conductivity of clay is designated kcl and effective conductivity of
sand+pebbles+water is designated as keff. Then, under steady-state, dependence of
th (or COP) and Tcold will be given by
R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 13

 , F , , , , H, , , , ,  ,  , , (3)

Figure 1. Clay pot refrigerator and assumed model.

where  is emissivity of the outer surface of the outer pot and i and 0 are heat
transfer coefficients associated with exposed surfaces of inner and outer pots.
The objective of the present paper is to assess this functional dependence
using the Reynolds Flow Model due to Spalding [3]. However, in [1], the treatment
of radiation heat transfer was not accounted in the manner prescribed in the
Reynolds flow model. As such, this corrected model is presented in the next
section. The computed results for several values of parameters are presented in
the subsequent section. Finally, conclusions of the analysis are reported.

REYNOLDS FLOW MODEL

Definitions of States and Phases

The Reynolds flow model is an algebraic model of mass transfer


(kg/s) across the interface between the t r a n s f e r r e d s u b s t a n c e
(water, in the present case) and the considered phase (stagnant surrounding air in
the present case). Figure 2 shows the mass transfer situation.
The interface is the outer surface of the outer pot  rw  r0  and is designated
by w-w. The width of the considered phase spans from w-w surface to the
imaginary - surface. In the Reynolds flow model, the transferred substance is
taken at uniform temperature and concentration. But, in the present case, the
temperature will vary radially in the annular space occupied by the transferred
substance. Therefore, we construct an imaginary transferred substance state M-M
with uniform (or, mixed-mean) properties given by
14 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

Ti  T0
TM  and v, M  1 (4)
2

where Ti and T0 are temperatures of the outer surface of the inner pot and inner
surface of the outer pot, respectively as shown in Fig 2. Note that, unlike in Ref
[1], the radiation heat transfer from the ambient Qrad (W) is accounted for as a
volumetric heat source in the transferred substance phase. However, the natural
convection heat transfer Qnc (W) is accounted at the interface between outer
surface and the considered phase as in [1]. Further, Q L (W) represents the
inward conduction heat transfer in the transferred substance phase. Finally, the
inner pot wall is assumed impervious to mass transfer but, permits heat transfer
Qload.

Steady-State Assumption

In reality, several factors shown in Eqn 3 vary with time. Variations of


ambient conditions T ,,  are obvious. However, time-dependence of quantity
such as Qload will depend on user-activity as well as the food. But, since the
objective of this paper is to identify main performance-influencing factors,
steady-state assumption is invoked. To maintain steady state, two assumptions
are made:

Figure 2. Reynolds Flow Model.


R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 15

1. Water is continuously injected into the annular space between the pots at
the rate that exactly equals the rate of evaporation ṁw.
2. The injected water temperature exactly equals imaginary mean temperature TM

The first assumption ensures that height (H) of water-column in the annular
space remains constant. The second assumption ensures that imaginary
temperature TM remains constant. Note that magnitudes of both ṁ w and TM are
apriori not known.

Reynolds Flow Hypothesis

The Reynolds Flow Model (see [3,4,2] and fig 2) postulates two mass flows.

1. A mass flow g A 0 (kg/s) crossing the  -  surface flowing towards the w-w
surface but carrying with it properties of the –state.
2. A mass flow (g A0 + ṁ w) crossing the  -  surface away from the w-w
surface but carrying with it properties of the w-state.

where Ao = 2 π ro H. Now, since ṁ w enters the considered phase at the w-w


surface, mass balance over the width of the considered phase shows that
hypothesizing existence of fictitious mass flow (g Ao) does not create or destroy
any mass in the considered phase. The magnitude of (g) is not affected by presence
of gradients of species mass-fractions j or temperature T, nor affected by
turbulence, radiation etc in the considered phase [2]. The g - flux is also not
affected by direction of ṁ w. With these postulates, we invoke conservation of
mass and energy principles.

Mass Conservation Principle2

Thus, invoking the mass conservation principle for water-vapour between w-w
and  -  states, we have

g A0 v,  m w   m w  g A0  v, w (5)

where  v is vapour mass-fraction. Of course, mass fraction of air will be


 a = 1 -  v . Upon rearrangement,

v,  v, w
m w, m  g A0  B where B  (6)
v, w  1

2
The developments in this subsection remain the same as in Ref [1]
16 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

and, since the Lewis number Le ≃ 1 (for our air-water vapour system), the mass
transfer coefficient g (kg/m2-s) can be estimated [3,4] from natural convection
heat transfer coefficient 0 as

(7)

In equation 6, subscript m designates that , is evaluated from mass-


conservation principle and B is called the Spalding number. The value of v, is
known but that of v,w is not known. The latter will be determined from
equilibrium condition at the w-w surface when Tw is known.
Further, this evaporation rate meets with resistance of the outer-pot clay wall
thickness. Therefore, using Darcy’s law, the average mass flow rate over height
H is

  w k p pa   pH /2 pw, sat 
m w, m       (8)
 wb0   pa pa 

where kp (m2) is permeability. Note that pH/2/pa = 1 + 0.05 H because pa  10 m of


water. Also, since saturation conditions prevail at the w-w surface, pw,sat/pa =
v,w/(0.622 + 0.378 v,w). Thus, we have

m w, m 0.622 1  0.05 H    0.0189 H  0.622  v, w


 (9)
 w khy A0 0.622  0.378 v, w

where khy = (kp pa) / (w b0) is hydraulic conductivity expressed in (m/s).

If we now add and subtract (0.0189 H – 0.622) v, in the numerator on the
right hand side of Equation 9, then replacing (v, - v,w) from Eqn 6, it can be
shown that

B
 0.622  0.0189 H  v,w  0.622 1  0.05H  (10)
 0.622  0.0189 H  v, w  1  Bim  0.622  0.378 v,w 


, (11)

and Bim  g/(w khy) may be viewed as mass transfer Biot number.
R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 17

Energy Conversion Principle3

With reference to Fig 2, writing the energy conservation principle for the
air-water vapour mixture between  -  and w-w surfaces, we have

g Ao h∞ + ṁw hM,w = Qnc + (ṁw + g Ao ) hw (12)

Similarly, energy balance across w-w and o-o surfaces (that is, thickness of the
outer pot), we have
Qnc ­ QL = ṁw (hM,w ­ hM,0) = ṁw C pw (Tw ­ To ) (13)

and, energy balance between M-M and o-o surfaces, can be written as

, T 0 T M
r r
or , (14)
or ,

Notice that Qo = -QL and QM = Qload. To make further progress, we substitute Eqn
13 in Eqn 12 to eliminate Qnc. Then, we have

, (15)

In the above equation, QL can now be replaced from Eqn 14. Algebraic
manipulations show that the evaporation rate , from the energy
considerations can now be expressed as


, (16)

This equation can also be cast in the form of Eq 6 and 11 as


(17)
/ ,

, (18)

3
Equations (12) to (18) are different from those in Ref [1]. They are derived by accounting
for Qnc and Qrad as per the provisions of the Reynolds Flow Model [3,4].
18 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

In the above derivations, following definitions are used.

   
h  Cpa T  Tref   Cpv  Cpa  T  Tref  ref  v,
 
(19)

   
hw  Cpa Tw  Tref   Cpv  Cpa  Tw  Tref  ref  v, w
 
(20)


hM , w  Cpw Tw  Tref  (21)

hM ,0  Cpw T0  Tref  (22)

Ti  T0

hM  Cpw TM  Tref  TM   2
(23)

where ref is latent heat of water at Tref. Equating estimates of evaporation rates
from mass (Eqn 11) and energy conservation (Eqn 16) principles provides for
iterative determination of unknown temperature Tw and evaporation rate m w . This
is explained in the solution-procedure sub-section.

Evaluation of T0, Ti and Tcold

Once Tw is known, the above temperatures are determined from following


relations.

Qload 1 bi 
Tcold  Ti     (24)
Ai 
 i k cl 

QL  Qload
Ti  T0  (25)
keff Ai / ( ri ln(r0 / ri ))

QL  Qload
T0  Tw  (26)
kcl A0 / b0

T0  Ti QL  Qload  Qrad
T0  TM   (from Eqns 14 and 23) (27)
2 m wCpw
R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 19

Inter-substitutions give following results:

1
 m w Cpw / 2 
QL  Qload  Qrad 1   (28)
 k A
eff i / ( ri ln ( r / r
0 i 
))

 b r ln(ro / ri ) 
Ti  Tw  (QL  Qload )  0  i  (29)
 kcl A0 keff Ai 

 b 
T0  Tw  (QL  Qload )  0  (30)
 kcl A0 

Qload  1 bi 
Tcold  Ti     (31)
Ai   i kcl 

Solution Procedure

Equations 11 and 16 are solved iteratively since the temperature Tw at the


outer surface of the outer pot is unknown. The main steps are

Preliminary Steps
1. Specify geometry parameters: ri,r0,bi,b0 and H
2. Specify ambient parameters T and  (or v,, see Appendix) and
respiratory load Qload.
3. Specify properties: Cpw, Cpa, Cpv,khy,ka,kcl and keff
Iterations Begin
4. Assume Tw and evaluate v,w from correlation give in Appendix.
5. Evaluate v,mean = 0.5 (v, + v,w). Hence, evaluate
Cpm  Cpa (1  v , mean )  Cpvv, mean .

6. Evaluate 0 from McAdams [5] correlation for natural convection from a


vertical surface NuH = 0.59 (GrH Pr)0.25 where
0H 9.81  (T  Tw ) H 3
Nu H  and GrH 
ka va2

Where  = 1/Tmean and Tmean = 0.5 (Tw + T) (K). Properties ka and va are
evaluated at Tmean
7. Knowing 0, evaluate g from Equation 7 and hence, Bim, B* and m w,m from
Equation 11
8. Evaluate Qin   Qrad  Qnc    rad   0  A0 T  Tw 
20 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

 
where  rad    T2  Tw2 T  Tw 
9. Evaluate h (Eqn 19) and hw (Eqn 20).
10. Evaluate QL + Qload (Eqn 28) where m w   m w,e  m w,m  / 2. Hence, evaluate Ti
(Eqn 29) and T0 (Eqn 30) and TM (Eqn 23)
11. Now, evaluate m w,e from Eqn 16
12. Calculate percentage difference
m w,e  m w, m
F x 100
m w, m

If F > 0.001, revise Tw to go to step 4.


13. Continue till convergence and evaluate COP from Eqn 1, Tcold from Eqn 31
and, hence, evaluate th from Eqn 2.

Solutions Tcold, th and COP for given T, , and Qload are of interest.

RESULTS AND DISCUSSION

Manner of Presentation

As noted under Preliminary Steps, the model requires several input


parameters. Among these, the most uncertain parameters are: (i) Hydraulic
conductivity khy (see [6,7]), (ii) Thermal conductivity of fired Clay kcl, (iii)
Effective thermal conductivity keff of sand+pebbles+water and (iv) Respiratory
load Qload (see [8]). In view of the uncertainties, we set-up a reference case with
values shown in Table 1.
The total cooling heat load comprising respiratory (3 kg food) + heat transfer
from top and bottom of the refrigerator ( 0.4 W ) is taken as Qload = 3 x 0.2 + 0.4 =
1 W.
Calculations are first presented for the reference case. Then, effects of
parameters are individually assessed keeping all other values fixed to those of the
reference case. In all calculations, the inner heat transfer coefficient is taken as i
= 0, Pr = 0.7 and emissivity  = 1.

Table 1. Parametric Values - Reference case.


R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 21

Results for Reference Case

Table 2 shows typical computed results for the reference case by way of
an example. The results show that Ti < To < Tw but Tcold > Ti as expected
permitting extraction of Qload across the inner pot surface. Also, the dew point
corresponding to ambient conditions Tdp < Tcold as expected giving th = 0.485.
The conduction heat transfer QL < 0 indicates that conduction in the transferred
substance phase is towards o-o surface4. Also, nc < rad indicating importance of
accounting for radiation heat transfer. Corresponding to Tw = 25.31990C, v,w =
0.02038 giving Spalding number B = 0.01609 which is further attenuated to B* =
0.00198 due to Darcy resistance. Note that Bh evaluated from Eq 17 equals B* as it
must; thus conforming excellent convergence. Finally, the mass transfer Biot
number Bim >> 1. Overall, except for sign of QL, all other results are in close
proximity of those presented in Ref [1].

Table 2. Results for the Reference Case:


T = 40oC,  = 10% (Tdp = 2.576oC), Qload = 1 W.

Parametric Variations

Effect of T and  : Table 3 shows the effect of relative humidity  at


3 values of T. It is seen that at each T, compared to dry ambient ( = 0), the
evaporation rate m w decreases whereas the outer pot surface temperature Tw
increases with increase in . Since (T-Tw) deceases with increase in  value of
nc decreases. Due to higher absolute temperatures, however, rad also increases.
The value of Spalding number B, as expected, decreases with increase in . But
that of B*, though smaller than B due to Darcy resistance, increases. The value of
COP increases with  mainly because of reduced Qin. Finally, the temperature
inside the inner pot Tcold increases with  indicating reduced cooling as expected.
Most importantly, Tcold > Tdp (see Appendix) in all cases. This observation thus
validates the main idea behind the definition of thermodynamic efficiency th. It is
seen that for a fixed T, th is maximum at   40 %. For each  , th increases
with T . Also note that at very high values of , th  0 indicating that Tcold > T
(see definition 2) and no cooling is achieved. This is again expected and,
therefore, does not invalidate the definition.

4
This is unlike in Ref [1] where QL was found to be positive because of inappropriate manner
of accounting of Qrad
22 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

Table 3. Effect of T and  - Qload = 1 W.

Effect of Qload : Table 4 shows the effect of assumed Qload on m w , th and
Tcold at T = 40oC by way of an example. Compared to the reference case of Qload =
1 W, for each , Tcold increases but th decreases with increase in Qload as
expected. Effect of Qload on m w is marginal.
Table 4. Effect of Qload at T = 40oC.
R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 23

Effect of keff : Table 5 shows the effect of thermal conductivity of sand +


pebbles + water keff. The table shows that compared to reference case of keff = 2
W/m-K, the values of Tcold are reduced for keff = 1 at each . At keff = 3, the
corresponding values of Tcold are higher. The table confirms our expectation that
value of keff should be as low as possible so as to enhance th. Effect of keff on m w is
again marginal.

Table 5. Effect of keff (W/m-K) at T = 40oC and Qload = 1 W.

Finally, as in Ref [1], the effects of variations in kcl and khy on the
performance were found to very marginal. Hence, detailed calculations are not
tabulated here.

CONCLUSIONS

In this paper, the steady state performance of a clay-pot refrigerator is


analysed using the Reynolds Flow model of mass transfer due to Spalding [3]. The
main conclusions are

1. For a given geometry, thermal and hydraulic conductivities and cooling load,
all parameters including COP, th and inner pot temperature Tcold show
expected magnitudes irrespective of ambient T and .
2. For a given T, COP and Tcold increase with increase in  but, Tcold > Tdp in
each case.
3. The validity of the newly defined performance parameter th has been
established through detailed computations. Irrespective of T, th,max occurs at
  40 %. Negative th implies no cooling.
4. The effect of thermal conductivity of sand+pebbles+water keff is most
pronounced. Lower value of keff is to be preferred to reduce Tcold and increase
th.
5. The effects of thermal conductivity kcl and hydraulic conductivity khy on Tcold
are found to be marginal.
24 R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25

REFERENCES

1. Date, A.W., 2012. Heat and Mass Transfer Analysis of a Clay-Pot


Refrigerator, International Journal of Heat and Mass Transfer, 55, 3977-
3983.
2. Date, A.W., Convective Heat and Mass Transfer, NPTEL Online Video
Course, http://www.nptelvideos.in/2012/12/convective-heat-and-mass-transfer.html,
Lecture 34.
3. Spalding, D.B., 1963. Convective Mass Transfer, Edward Arnold Ltd.,
London.
4. Crawford, M., and Kays, W.M., 1993. Convective Heat and Mass Transfer,
McGraw-Hill International Edition, New York.
5. McAdams, W.H., 1954. Heat Transmission, 3rd Edition, McGraw-Hill, New
York.
6. Abu-ZreigM, M., and Atoun, M.F., 2004. Hydraulic Characteristics and
Seepage Modelling of Clay Pitchers Produced in Jordan, Journal of Canadian
Biosystems Engineering, 46, 1.15-1.20.
7. Gil, M., Sinobas, L.R., and Juana, L., 2010. Evolution of Spherical Cavity
Radius Generated Around a Subsurface Drip Emitter, Biosciences Discussion,
7, 1935-1958.
8. Cengel, Y.A., and Ghajjar, A.J., 2006. Heat and Mass Transfer:
Fundamentals and Applications, 4th Ed, McGraw-Hill, Ryerson.
9. Wagner, W., and Pruss, A., 1993. International Equations for the Saturation
Properties of Ordinary Water Substance. Revised According to the
International Temperature Scale of 1990, Journal of Physical Chemistry. Ref.
Data, 22, 783-787.
R.M. Damle, A.W. Date / Journal of Energy, Heat and Mass Transfer 37 (2015) 11-25 25

APPENDIX

1. Knowing Tw, v,w is evaluated from [4]

  
v, w  3.416  103  2.7308  104 Tw  1.372  105 Tw2 

 8.2516  10    6.9092  10 
8
Tw3 9
Tw4

  3.5313  10  T   3.7037  10  T
10 5
w
12 6
w

  6.1923  10  T   9.9349  10  T
15 7
w
17 8
w

Correlation is valid for – 20 < Tw (C) < 100.

2. Knowing relative humidity , v, is evaluated from [9]

W  pv, 
v,  where W  0.622    and
1  W p  p 
 tot v , 
pv,   psat  F  T
   and = exp   where   1 
psat  100  pcr 1    Tcr

F  a1    a 2   1.5  a3   3  a 4   3.5  a5   4  a6   7.5

a1 = -7.85951783, a2 = 1.84408295, a3 = -11.7866497,


a4 = 22.6807411, a5 = 15.9618719, a6 = 1.80122502

Here Tcr = 647.096 K, pcr = 220.64 bar and ptot = 1.01324 bar

3. Dew Point temperature Tdp may be evaluated from [9]

237.7   17.271  T  
Tdp  where    ln   
17.271   237.7  T  100 

You might also like